Commentary and Additional Information

Technical Details

Fraxplorer and Fraxflame images were created and rendered on a Mactintosh. Over the years I have used various machines, from a 9600 200MP running MacOS 8.6 to a dual-G4 800MHz running OSX. Up until mid-2002 when I purchased a PIII 400MHz, all Ultra Fractal images were also created and rendered on various Macintoshes under Windows emulation (surprisingly, this worked rather well, although rendering took longer than optimal). In August 2004 I upgraded my PC to a P4 3GHz with 1.5GB of RAM in order to use Apophysis and do more deep Mandelbot zooms in Ultra Fractal, and in June 2006 I acquired a second PC (dualcore 3.2GHz Intel processor with 2GB of RAM) to act as a secondary render box. In December 2007 I acquired an Intel MacBook Pro, and am now using that machine for all creation work with all programs. Windows applications are running in a virtual environment under VMWare Fusion; Mac OSX apps are running natively. Previously mentioned Intel machines running Windows are now used as render boxes only.

Most images on this site were rendered out at 3200x2400 pixels (4:3) or 3200x2000 pixels (widescreen) or larger, and then reduced to a size more suited to the web. Rendering times varied from a few minutes to several days depending on the program, fractal type and complexity, however most images took 30 minutes to five hours to render.

This site is now assembled on a MacBook Pro running OSX using Photoshop to prepare the images, vi (a UNIX text editor) to write the HTML and iCab (a Mac-only web browser) to test the site and identify errors.

This site is served from a Debian Linux server running Apache2.

Some Background...

I first saw a fractal when I was about 12 years old. My father had taken me to a lecture on fractals at the local university, and I was completely fascinated by the images that accompanied the lecture. Of course I understood little about the math behind the images, but the imagery stayed in my memory until I got a computer in 1991 and acquired a copy of a free software fractal generator called Fractint. I quickly filled up all my available disk space with images and had to stop.

I got back into fractals in 1999 with the advent of a program called KPT Fraxplorer, which operates as a plug-in for Photoshop (or any other program that can use Photoshop plug-ins). In 1999 I had both a bigger hard drive, and a CD burner, so my creativity was not limited by my computer's resources. In 2000, I discovered Ultra Fractal, and in 2001 I started working with KPT Fraxflame, which I'd had since 1999, however I found it slow and awkward to run on the computer I had in 1999.

All of the images on this site are created because of my love of the beauty of fractals. In a practical sense however, many of these images are used as desktop backgrounds for myself and my friends. Knowing that an image is going to cover the background of my computer screen tends to lead towards images that are less focused on the centre - a location that is usually covered by whatever windows I am working with on the computer. Instead, the images tend to spread themselves out across the canvas.


I'm always surprised at how few people on the Internet display images created in KPT Fraxplorer. It is included in the same software package as KPT Fraxflame, which seemed to enjoy great popularity (at least until Apophysis came along). I can only conclude that it is somehow unpopular. It is true that it is a very simple program, and sometimes I compare the process of working in Fraxplorer to doodling. There are not a lot of options in terms of formulas or colouring methods, however I feel that this is its strength. Within every fractal, there is an infinite number of variations waiting to be explored; one merely needs some time and patience to do this. The images created with Fraxplorer are generally much simpler than images from many other programs. There is only one layer to work with, and no textures can be applied. The lack of these features allows me to concentrate on the structure of the fractal, and also on the colour. The result is a simpler, more basic image, but one that catches my eye nonetheless. Take for example this image: Grace. It is comprised of a single layer with only the basic colouring method applied, however it is my favorite desktop image and remained as my background for more than 6 months (most images last less than a month). I liked this image so much that I rendered a large-scale (30"x50" at 300dpi) version with the idea of having it printed as a poster. See a reduced size copy of the poster-format image here (720x1280 pixels, 400K). This image benefits greatly from additional resolution because the fractal is not completely self-similar and there is much of interest to see in the detail.

Initially I created mostly 'one-off' images. Nowadays, I'm creating a lot of image series - a number of images closely related by fractal formula and/or colour. This proliferation has lead to some 2300+ images to date. Obviously, not all of these are good, and more recently, many of them very are similar to each other, so only the best one or two of any series are displayed here. Every once in a while I think that I have done everything that can be done with Fraxplorer, but time and time again I prove myself wrong.


I didn't start working with Fraxflame until 2001, mostly because it is very processor intensive and I didn't have a machine that could easily keep up with it until then. Trying to learn what a new program can do when it takes minutes to get feedback from every little change can be very frustrating. The original Fraxflame program isn't the easiest program to work with. A number of obvious features seem to be missing, and the operation of the program is different than that of most fractal programs. In other fractal programs, the user chooses a formula, modifies the formula by giving the program different values when it calculates reiterations, and then (optionally) applies a colouring method. Fraxflame operates more like controlled evolution. The user is presented with a random selection of fractals when the program is opened. You chose an image, then select the level of "genetic mutation" desired, and then have the program regenerate another selection of fractals based on the chosen fractal. There are no colouring methods in Fraxflame, although you can choose some basic types of flame fractals to work with. If you chose a high level of mutation, the resulting fractals will not closely resemble their "parent". If you set the level too low, they will all look identical to their parent. Some of the images I have created have gone through more than 50 mutations between the first 'interesting shape' that randomly appeared and the final image. The problem with having random images pop up every time the program is opened becomes apparent when I find myself busily saving all the neat images that suddenly appeared, when I really should be rendering out the images I've already completed. Fraxflame II is a bit of an improvement as there are more ways to control how the fractal mutates and changes, however I still cannot do simple things like rotate the fractal on the screen before rendering. As of March 2004, most of the Fraxflame images on this site were made with the original Fraxflame.

The way gradients appear in Fraxflame is radically different than in Fraxplorer. Both programs are part of the same software package, so they use the same format to save gradients. However, what looks like a nice gradient in Fraxplorer may have serious problems when applied to a Fraxflame image. It may produce an image that is highly oversaturated, or washed out, or too bright. I find that I spend a lot of time adjusting gradients to look good in Fraxflame.

The other major problems with Fraxflame are the long rendering times and the wildly variable results that occur between rendering an image at 640x480 and 1600x1200 or larger. The larger the image is rendered at, the darker the final image usually is. Sometimes this can be corrected using contrast/levels/curves in Photoshop, but other times a re-render with the intensity level set higher is required. Unfortunately there's no foolproof way to tell what the results will be until the render is completed. And last but not least, the program has a render-size limit. Many images will not render larger than 2048x1536; and in fact, the program does not give an error when the render is started, but sits there for several hours or more looking like it's making progress before finally crapping out and failing. This is why most Fraxflame images are not available at sizes larger than 1600x1200 and 1920x1200. A few images that aren't very "zoomed in", such as Flight and Phoenix were able to be rendered at 2560x1600, but the rest of the images that I tried to render for that size failed. Given the problems I have getting images to render at desktop size, I don't ever expect to be able to print these images at any size larger than a postcard.

All that being said, the images that can be produced are very unusual and beautiful and the results are generally worth the effort. However, now that there are better fractal flame generators available, such as Apophysis, I don't use Fraxflame much anymore.

Ultra Fractal

Ultra Fractal is truly a giant leap forward in fractal software technology. I can't say thank you enough to Frederik Slijkerman (the creator of Ultra Fractal) for writing such a fantastic program and for making it available at such a reasonable price. Also deserving of thanks are all the people who have written fractal formulas and made them available for free on the Internet. This has enabled Ultra Fractal to be somewhat of a community project, and I think everyone benefits by this.

Ultra Fractal has (in my opinion) the distinction of being one of the most simple programs to use, yet it contains immense complexity that can take years to master. My own learning curve for Ultra Fractal took about two years of on-and-off work to complete to the point where I was capable of consistently creating the images I wanted to be making.

Just before I discovered Ultra Fractal, I was looking for ways to do more with Fraxplorer using Photoshop's layering abilities. While the idea was a good one, the method was somewhat clunky and resulted in a lot of extra work. Then I discovered that Ultra Fractal had these features built-in, allowing me to see the final image before it was rendered out. Adding the ability to layer different images on top of each other opens up endless possibilities (as if they weren't endless to begin with). Basic fractal shapes can be enhanced, textures can be added and an image can be more fully crafted, rather than just 'found'. Often the individual layers look nothing like the final image, especially with regards to colour. One of the biggest challenges to working with multi-layer images is controlling the colour. Many times I have created an image with a shape I like, but I am unable to adjust the colours to be something pleasing without ruining the shape of the image. This balances out somewhat, however, because colour palettes that I would never have created on my own sometimes appear in the course of working on an image. Ultra Fractal also requires a fair amount of patience; most images take several hours of work and some images have only come at the end of a night-long fractal marathon. As I become more proficient with the program, I'm starting to make images that require several days of continually building one image from another to end up with the final result. While I have occasionally made some really good images in a shorter period of time, it is generally not a program to spend ten minutes with and expect to create anything decent.

And unlike Fraxplorer and Fraxflame, Ultra Fractal does not hide the math from the user. While I make no claim whatsoever to having even a vague understanding of the math that goes into fractals, it's nice to have more options to twiddle with, and is also useful for explaining to more mathematically inclined friends how an image was made.


In August 2004 I started working with Apophysis, a GPL licensed fractal flame generator for Windows. This finally prompted an upgrade in my PC, from a PIII 400MHz to a P4 3GHz with 1.5GB of RAM. Apophysis may be a resource hog, however I'm very pleased with the operation and results from it. Compared to Fraxflame, Apophysis is a far superior program in many ways, including number of flame types, options in manipulating the flame, the ability to render one image while still continuing to work, the ability to import saved flames into Ultra Fractal, and WYSIWYG rendering, even for large images (density compensation). Apophysis unfortunately only has a primitive gradient editor, however cut and paste from Ultra Fractal's excellent gradient editor works well.

Apophysis also does not seem to have the same render-size limitations that exist in Fraxflame. Rendering poster-sized images does take a long time, but it is now possible to do, whereas Fraxflame sometimes had difficulty rendering images for large desktops (1600x1200 and larger). Apophysis has no problems rendering at large sizes and now by default I render all images at 3200x2400. The final render requires no adjustment and looks far superior in quality and detail, unlike Fraxflame where the image generally comes out dark or faint and requires major adjustments in brightness, contrast and saturation, or a re-render (or both) to make it look like the 640x480 version. This is a really nice change. I highly recommend this program for anyone with a Windows machine who is interested in experimenting with flame fractals.

The recent release of Apophysis v2.03c has brought a number of new features, including new flame types (fan2 is my current favorite), better control over transforms, the ability to lock in the preview window size (which allows for more consistent results when working with odd-shaped images), and finally, the ability to rotate flames prior to rendering.

Addendum: I stopped using Apophysis in April 2006 after version 2.03c expired on me. I hadn't realized it was a beta version, and found out the hard way. Since then, all new releases have been beta versions which expire. As an artist, I need to be able to reproduce my work at a later time; at a larger size, better quality, or different ratio, for both my backgrounds site and for commissions. Playing jiggery-pokery with the clock on my computer in order to reproduce works that I rely up on for my income is not acceptable. So I have abandoned Apophysis until a stable (non-expiring) version comes out, or I have time to track down a copy without an expiry date. While Apophysis is GPL and the source code is available, unfortunately I have neither the software (Borland Delphi), or the programming skills to do this myself.

This is doubly unfortunate, because the newer versions of Apophysis can do some truly amazing things now that were not possible before. However overall, the images I made in Apophysis are still some of the most popular works on my site, so despite their lack of technical sophistication by today's standards, I will continue to post them as backgrounds.

Addendum 2 (January 2007): I have re-started work with Apophysis thanks to a variant version of Apophysis released by Joel Faber, who has released it without an expiry date. You can see some of my new Apophysis images in Apophysis Gallery 19 and Apophysis Gallery 20.